Feasibility study of amperometric and electrochemical quartz crystal microbalance measurements for in situ state of charge monitoring in vanadium flow batteries

For an efficient flow battery operation, knowledge of the state of charge of the battery is essential. Monitoring the state of charge of both half cells is advantageous concerning a timely detection of crossover processes. We present the first results for amperometric and electrochemical quartz crystal microbalance measurements in a vanadium flow battery test setup. By validating with half cell potential measurements as well as ex situ titration we investigated the applicability of both electrochemical methods for an in situ half cell state of charge monitoring.


Introduction
To overcome the challenges to mitigate climate change, the storage of renewable energies is essential.Flow batteries (FB) are most suitable to store wind or solar energy due to the characteristic independent scalability of energy and capacity [1,2].Of all FB chemistries, all-vanadium flow batteries (VFB) are most common and commercially prevailed.They contain vanadium electrolyte in the negative half cell (NHC) as well as the positive half cell (PHC) [3].The charging (from left to right) and discharging (from right to left) processes of the NHC and PHC are characterised by reactions (1) and (2): V 3+ + e -⇄ V 2+ (NHC) (1) VO 2+ + H2O ⇄ VO2 + + e -+ 2H + (PHC) (2) During charging, V(III) ions are reduced to V(II) in the NHC while the oxidation of V(IV) ions to V(V) occurs in the PHC.During discharging V(III) ions and V(IV) ions are formed in the NHC and PHC, respectively.The crossing of vanadium species through the ion exchange membrane, which separates the PHC from the NHC, does not lead to a contamination of the half cell electrolytes since both half cells contain vanadium.However, the battery half cells are suffering from self discharge.Furthermore, the half cell electrolytes may undergo other reversible and also irreversible changes as e.g.concentration changes due to osmotic processes, overoxidation and electrolysis of the electrolyte [4][5][6][7][8].Therefore, knowledge of the state of charge (SOC) of the battery is necessary for an efficient operation of the VFB.In commercial VFB the so called reference cell monitoring is widely used [4][5][6][7][8][9] but this monitoring method is limited to the determination of the SOC of the full cell electrolyte and does not give information about electrolyte crossover and self discharge of the battery half cells.
Monitoring SOC of each half cell separately is much more suitable than reference cell monitoring for a timely detection of crossover and other processes which alter the half cell electrolytes.Since, next to oxidation state of the electroactive vanadium species, several electrolyte properties such as concentration, conductivity, pH, density, viscosity and absorbance may change during charge and discharge, numerous measurement methods are suitable [1] for an in situ detection of these changes.Monitoring the redox potential of the half cells using reference electrodes [4,5,[9][10][11] as well as using UV/Vis spectroscopy [12][13][14][15] to monitor the oxidation state and concentration of the vanadium ions the half cell electrolytes have been widely investigated.In addition to that, density measurements [10], viscosity measurements [16] and distribution of relaxation times (DRT) analysis applied on electrochemical impedance spectroscopy (EIS) data [17] have been implemented to investigate the VFB electrolyte in situ.
Amperometric measurements might be suitable for a novel in situ half cell monitoring in VFB.This method has already been studied for ex situ and in situ monitoring using microelectrodes in organic FB [3,[18][19][20].Various potentials have been applied and the ratio of resulting oxidation and reduction currents has been plotted vs. the ratio of the concentration of the oxidized and reduced species in the electrolyte.An outstanding advantage of this approach is the independence on temperature, electrolyte composition and concentration [18].For VFB, the suitability of amperometric measurements using working electrodes coated with porous diffusion layers for an ex situ determination of the SOC, has been shown [2] and also an electro catalytic metal electrode for estimating the SOC of the positive electrolyte has been patented [21].
Also, the electrochemical quartz crystal microbalance (EQCM) might be suitable for SOC monitoring.An EQCM consists of a quartz crystal in contact with electrodes on both sides of the quartz which stimulate oscillation.The electrode facing the electrolyte is also contacted as working electrode to a potentiostat.Simultaneously with the changes in oscillation frequency of the quartz crystal, electrochemical reactions can be analysed at the electrode [22].EQCM has been already used for density and viscosity measurements [23,24] and battery monitoring in general [25] as well as for measuring density and viscosity in lead acid batteries in particular [26].EQCM therefore provides the possibility to investigate two different electrolyte parameters as redox potential and viscosity at the same time within the same setup [25].
We are investigating these two different electrochemical methods according to their feasibility for SOC monitoring in VFB.

Experimental
The battery test setup for this study contains the battery cell and a tank for each battery half cell.The electrolyte is pumped with 1.5 -2.1 l/h electrolyte flow using a membrane pump (ProMinent, delta opto drive).The battery cell with an electrode area of 10 cm 2 has been manufactured by the machine shop at the DECHEMA-Forschungsinstitut and is equipped with a F10100 membrane (Fumatech BWT GmbH), GFD 4.6 EA electrode felts (SGL Carbon) and SP4369 bipolar plates (Schunk Kohlenstofftechnik GmbH).Commercial vanadium electrolyte (GfE Metalle und Materialien GmbH) with a vanadium concentration of 1.6 M and acid concentration of 2 M is used.For charge and discharge experiments volumes between 60 and 150 ml electrolyte are filled in each tank.The battery cycling is performed under argon with charge/discharge currents of 50 or 100 mA cm -2 and voltage criteria of +1.65V (charged) and +0.8 V (discharged) using a Gamry Reference 3000AE Potentiostat/Galvanostat.The components and parameters described have been utilized both for the amperometric and EQCM measurements.Additionally, each half cell electrolyte circuit is equipped with a flow cell to measure open circuit potential (OCP) of the half cell electrolyte.Both flow cells are connected by an electrolyte bypass to the battery test setup and the electrolyte is pumped through the bypass using a ISM796 peristaltic pump (Ismatec).A glassy carbon (GC) rod with 2 mm diameter (Sigradur D, Hochtemperatur-Werkstoffe GmbH) and a Hg/Hg2SO4 reference electrode (Xylem Analytics Germany Sales GmbH & Co. KG) in a Luggin capillary (filled with 2 M H2SO4) are mounted in each flow cell to measure OCP. Figure 1 provides a scheme for the test set up with cells for OCP (insertion in the middle), amperometric (insertion right) and EQCM (insertion left) measurements.The electrochemical cells used for amperometric and EQCM measurements are described below.

Figure 1. Scheme for VFB test setup with electrochemical cells for in situ amperometric measuerements in the positive half cell (PHC) and in situ electrochemical quartz crystal microbalance (EQCM) measurements in the negative half cell (NHC). Open circuit potential (OCP) measurements. Cells equipped with electrodes: working (WE), reference (RE) and counter (CE)
Ex situ calibration of the mentioned monitoring methods has been done using potentiodynamic titration (Titrando 888, Deutsche METROHM GmbH & Co. KG) with 0.1 M Ce(SO4)2 (Carl Roth GmbH & Co. KG).As sensor, a platinum indicator electrode has been utilized for all measurements (Combined iPt ring electrode, Deutsche METROHM GmbH & Co. KG).The equivalence points have been calculated with Tiamo v2.5 (Deutsche METROHM GmbH & Co. KG).
The concentration of V 4+ has been determined by titration using Ce(SO4)2.Afterwards FeSO4heptahydrate has been added until the potential dropped below 550 mV relative to the sensor

CE WE RE WE RE CE RE
WE Quartz electrode to ensure that V 5+ is reduced quantitatively.Therewith, the total concentration of vanadium in the positive electroyte has been titrated using Ce(SO4)2.V 5+ concentration has been calculated as the difference of the total vanadium concentration and V 4+ concentration.The SOC has been determined as the proportion of V 5+ on the total vanadium concentration.
Amperometric measurements are tested (using a Gamry Reference 3000AE Potentiostat/Galvanostat) concerning their feasibility for SOC monitoring at VFB and several setup configurations have been applied: A. Ex situ three electrode setup using a carbon microelectrode as working electrode (carbon fibre with 33µm diameter, MCE Micro Carbon fiber electrode, ALS Co., Ltd.), Hg/Hg2SO4 reference electrode (Xylem Analytics Germany Sales GmbH & Co. KG) in a Luggin capillary (filled with 2 M H2SO4) and a glassy carbon rod counter electrode with 2 mm diameter (Sigradur D, Hochtemperatur-Werkstoffe GmbH), all electrodes installed in a small glass cell (20 ml cell volume).The battery has been operated with 100 ml electrolyte in the NHC tank and 60 ml in the PHC tank to ensure full charging and discharging of the PHC electrolyte.During battery charge and discharge at several SOC, 1 ml electrolyte has been withdrawn from the tank of the PHC, added to the amperometric cell subsequently mixed with 2 ml ultrapure H2O.Another 0.5 ml electrolyte has been withdrawn from the tank to determine the SOC by titration (see paragraph above).For the camperometric measurement, a reduction potential of (-1.2 V vs. Hg/Hg2SO4) has been applied to the microelectrode for 30 s and then an oxidation potential of (+1.0 V vs. Hg/Hg2SO4) has been applied for 30 s. Reduction and oxidation have been applied in an alternating manner several times (see Fig. 4).All amperometric measurements have been conducted under argon protection.B. In situ three electrode setup using a carbon microelectrode as working electrode (carbon fiber with 33 µm diameter, ALS Co.,Ltd.),Hg/Hg2SO4 reference electrode (Xylem Analytics Germany Sales GmbH & Co. KG) in a Luggin capillary (filled with 2 M H2SO4) and a glassy carbon rod as counter electrode with 2 mm diameter (Sigradur D, Hochtemperatur-Werkstoffe GmbH).The amperometric cell has been connected to the electrolyte bypass with two hoses, one for filling and one for emptying the amperometric cell.In order to carry out the amperometric measurement, the cell has been filled up to a level which was constant for all measurements (7 ml) and convection has been avoided.A reduction potential of (-1.2 V vs. Hg/Hg2SO4) has been applied for 30 s and an oxidation potential of (+1.0 V vs. Hg/Hg2SO4) has been applied for 30 s.
After the amperometric measurement, 0.5 ml electrolyte has been withdrawn from the cell for SOC determination via titration.Afterwards, the electrolyte in the amperometric cell has been pumped back through the bypass to the electrolyte circuit.This "stop and flow" methode for electrolyte sampling in the amperometric cell has also been used for setup C and D. C. In situ three electrode setup using a glassy carbon rod as working electrode with 2 mm diameter (Sigradur D, Hochtemperatur-Werkstoffe GmbH), Hg/Hg2SO4 reference electrode (Xylem Analytics Germany Sales GmbH & Co. KG) in a Luggin capillary (filled with 2 M H2SO4) and a glassy carbon rod counter electrode with 2 mm diameter (Sigradur D, Hochtemperatur-Werkstoffe GmbH).The setup and the measurement are described in B. A reduction potential of (-0,7 V vs. Hg/Hg2SO4) has been applied for 30 s and an oxidation potential of (+0.9 V vs. Hg/Hg2SO4) has been applied for 30 s. D. In situ two electrode setup using a glassy carbon rod as working electrode and a second glassy carbon rod counter electrode with 2 mm diameter (Sigradur D, Hochtemperatur-Werkstoffe GmbH).The setup and the measurement are described in B. A reduction potential of (-1.0 V) has been applied for 30 s and an oxidation potential of (+0.7 V) has been applied for 30 s.The Electrochemical quartz crystal microbalance (EQCM) consists of a commercially available cell case (Deutsche METROHM GmbH & Co. KG) in which the quartz crystal is inserted.A carbon plate has been installed as a counter electrode (CE).The cell head has been manufactured to work as a flow cell.Electrolyte inlet and outlet and the Luggin capillary with Hg/Hg2SO4 reference electrode (Xylem Analytics Germany Sales GmbH & Co. KG) have been connected to the cell head.After assembling the cell, it has been connected to the electrolyte bypass circuit and filled from the bottom and emptied from the top in order to avoid accumulation of bubbles on the electrodes.The EQCM cell has been connected to the bypass electrolyte of the NHC and the oscillation frequency of the crystal and the potential between the carbon coated crystal and Hg/Hg2SO4 RE (OCP) have been monitored during battery cycling.For controlling the EQCM experiment, a PGSTAT302N Potentiostat (Deutsche METROHM GmbH & Co. KG) and gold coated quartz crystals with a thin layer of carbon evaporated onto one side (6 MHz resonance frequency, RenLux Crystals Ltd.) serving as working electrode in the EQCM cell have been utilised.Carbon has been applied as coating material to mimic the carbon electrodes used in VFB.
The battery cell has been cycled between 0.8 and 1.65 V with a current density of 50 mA cm -2 during the EQCM experiments.

Results and discussion
The VFB in the test setup has been charged and discharged using the parameters described above.Figure 2 shows exemplary operation of the battery cell respectively the resulting potentials during charging and discharging with 50 mA cm -2 .The VFB is running stable, and the charge/discharge cycles are reproducible.Since the battery has been operated with an excess of NHC electrolyte to ensure full charging and discharging of the PHC electrolyte, a slightly increasing duration of the cycles can be observed which is due to crossover of NHC electrolyte through the cation exchange membrane and therefore capacity increase in the PHC.During cycling of the battery cell, the open circuit potential (OCP) has been monitored for both half cells using flow cells equipped each with a GC rod and a Hg/Hg2SO4 reference electrode.Figure 3 shows the SOC calculated from OCP measured in the positive half cell (PHC) (grey curve) in comparison to the SOC calculated from the titration of samples taken from PHC electrolyte during the amperometric measurements.The SOC calculated from in situ OCP measurements in the PHC is in good accordance with the SOC determined by titration (black crosses).

Figure 3. State of charge (SOC) calculated from open circuit potential (OCP) measurements in the positive half cell (PHC) and SOC calculated from titration (black crosses)
The titration samples have been taken from the PHC and analysed directly after several amperometric meaurements.The results for these measurements are described below and are shown in Figsures 4 and 5.

Amperometric measurements
At first, the three electrode setup using a carbon microelectrode (A) has been tested ex situ to ensure a convection-free measurement at the microelectrode.Figure 4 shows that for the amperometric experiments the reduction current resulting from polarisation with -1.2 V increases with increasing SOC whereas the oxidation current resulting from polarisation with +1.0 V decreases with increasing SOC.Since in the PHC the share of vanadium ions in the charged state (V 5+ ) increases with SOC, the reduction current increases because the amount of reducible vanadium ions increase.Vice versa, the oxidation current decreases with increasing SOC since the amount of oxidisable V 4+ ions decrease.The currents measured for the reduction and oxidation steps are reproducible.The average of the last 10 seconds of the measured currents is referred to as oxidation/reduction current and the ratio ln(Iox/Ired) is calculated according to [3,18].
The ln for the ratio of the measured oxidation and reduction current is supposed to linearly depend on the ln for the concentration of the oxidised and reduced vanadium ions in the electrolyte [18].According to theory, an equal share of oxidised and reduced species (ln(cox/cred) = 0) should result in equal oxidation and reduction currents (ln(Iox/Ired) = 0).Plotting the raw data, it has been observed that all data points are shifted towards higher ln(cox/cred) values, which has also been observed for all amperometric measurements.Because of that, all measured currents have been corrected by the y-intercept of the linear regressions calculated for each set of oxidative and reductive currents, respectively, according to [18].Nevertheless, the slope of the regression line is expected to be -1 for all experiments.
Within the following the ln(Iox/Ired) vs. ln(cox/cred) plots for the corrected currents are shown and discussed.The plot of ln(Iox/Ired) as determined by amperometric measurements vs. ln(cox/cred) as determined by titration (as explained in the Experimental section) is shown in Figure 5.For easier understanding, the SOC values calculated from titration results are added.To investigate the applicability of this setup for in situ monitoring, the amperometric cell was connected to the PHC electrolyte circuit and operated as described in B.
In Figure 6, the results for the in situ three electrode setup using a carbon microelectrode (B) are shown and the SOC calculated from titration is added.The results are in good accordance to the results for the ex situ three electrode setup using the microelectrode.The regression line again shows a slope of nearly -1.
Since a microelectrode might be susceptible for interferences when applied in a technical VFB system, an in situ three electrode setup using a GC rod (C) working electrode has been tested in order to simplify and to optimise the amperometric setup regarding maintenance and cost efficiency.
Figure 7 shows that, using a GC rod working electrode, ln(Iox/Ired) also results in a linear dependency on ln(cox/cred) with a slope of nearly -1.Comparing Figures 4, 5 and 6, differences in the scaling of the axis for ln(Iox/Ired) and ln(cox/cred) are due to different SOC ranges in which the experiments have been done and samples have been taken.For further simplification and cost reduction, the reference electrode has been omitted and the results for amperometric measurements with an in situ two electrode setup using a GC rod (D) working electrode are shown in Figure 9.The results reveal that also for this simplified setup the results are in good accordance to the results gained for the three electrode setup using a microelectrode.The results for ln(Iox/Ired) still show a linear dependency on ln(cox/cred) but with a slope which deviates slightly stronger from -1.This might be caused by a potential shift at the glassy carbon rods during the experiment.In addition, we also have been testing the electrochemical quartz crystal microbalance (EQCM) concerning its applicability for SOC monitoring.The quartz crystal coated with a carbon layer as working electrode on one side can be utilized to measure OCP in the electrolyte circuit and furthermore measuring the oscillation frequency may allow the investigation of electrolyte properties as density and viscosity at the same time respectively within the same setup.To check the applicability, an EQCM cell has been mounted in a bypass of the NHC electrolyte circuit of the VFB test setup and the battery cell has been charged and discharged.Figure 10 (a) shows that the results for OCP measured at the carbon coated quartz crystal and OCP measured at the GC rod in the flow cell are comparable and in good accordance.Figure 10 (b) shows the frequency of the crystal which has been measured simultaneously and also in a currentless state.The results reveal that changes of the frequency of the oscillating quartz crystal depend on the SOC of the electrolyte.High frequencies which are expected to be measured for low electrolyte densities are observed for high OCP respectively high SOC of the NHC.This is in good accordance to results published for VFB electrolyte densities in dependence on SOC [10].For the charged state of the NHC, the density of the electrolyte is low which results in high frequencies whereas the higher electrolyte density for the discharged negative half cell electrolyte results in lower frequencies.The frequency shift from cycle to cycle, which can be observed predominantly for the charged NHC, might indicate crossover processes.Since the method is temperature dependent, temperature has been also monitored and plotted in Figure 10 (b) but the observed frequency changes are not caused by changes in the electrolyte temperature.
The feasibility of EQCM measurements to identify crossover of different electrolyte components (active vanadium species, sulfuric acid, water) will be investigated further.

Conclusions
Based on our first results for amperometric measurements in the electrolyte of a positive half cell of a vanadium flow battery laboratory test set up, we conclude this method is applicable for the in situ determination of the state of charge (SOC) of the positive half cell (PHC) electrolyte in vanadium flow batteries.A linear dependency of the current ratio (ratio of the measured oxidation and reduction currents, ln(Iox/Ired)) on the concentration ratio (ratio of the concentration of the oxidised and reduced vanadium ions, ln(cox/cred)) in the PHC has been found for several test setups and electrodes.A three electrode setup using a carbon microelectrode as well as a glassy carbon (GC) rod and also a simplified two electrode setup using two GC rods have been successfully applied for state of charge (SOC) monitoring in the PHC electrolyte.Additionally, SOC monitoring of negative half cell electrolytes using amperometric measurements will be investigated in the future.For all tested setups the choice of suitable oxidation and reduction potentials is essential for an accurate SOC determination.In addition to that, the ratio of diffusion coefficients and the long-term stability of the electrodes will also be considered in future research.Since all shown results have been gained in situ but with a stop and flow sampling of the electrolyte to ensure convection free amperometric measurements, we will further investigate and develop a flow cell and a polarisation routine which allows an amperometric monitoring of the SOC in a flow-through mode.Since this method is simple concerning technical equipment and maintenance, we predict an applicability for field use in commercial vanadium flow batteries.
Summarizing the electrochemical quartz crystal microbalance (EQCM) results, we conclude that the redox potential respectively the open circuit potential of the negative half cell electrolyte can be monitored using a carbon coated quartz crystal.Furthermore, the oscillation frequency of this quartz crystal depends on the state of charge (SOC) of the electrolyte and therewith offers a new opportunity for SOC monitoring.As far as we know, an application of EQCM measurements for in situ SOC monitoring has not been published yet and we will further investigate the suitability of this technique for SOC determination as well as crossover identification and monitoring of different electrolyte components for the negative as well as the positive half cell of vanadium flow batteries.Also, the sensitivity to changes in temperature and concentration of the electrolyte will be investigated as well as the stability of the electrodes.Due to the complexity of the EQCM setup and its susceptibility for interferences we propose rather an application for laboratory research than for field use.

Figure 4 .
Figure 4. Ex situ amperometric measurements at the positive half cell with three electrode setup using a microelectrode

Figure 5 .Figure 5
Figure 5. ln(Iox/Ired) vs. ln(cox/cred) plot for ex situ amperometric measurements at the positive half cell with three-electrode setup using a microelectrode

Figure 6 .Figure 7 .
Figure 6.ln(Iox/Ired) vs. ln(cox/cred) plot for in situ amperometric measurements at the positive half cell with three-electrode setup using a microelectrode

Figure 8 .
Figure 8. Iox/Ired vs. cox/cred plot for in situ situ amperometric measurements at the positive half cell with three-electrode setup using a glassy carbon rod

Figure 9 .
Figure 9. ln(Iox/Ired) vs. ln(cox/cred) plot for in situ amperometric measurements at the positive half cell with 2-electrode setup using glassy carbon rods

Figure 10 .
Figure 10.(a) Negative half cell: open circuit potential (OCP) measured in an EQCM cell compared to OCP measured in a flow cell at a glassy carbon rod.(b) Oscillation frequency of the quartz crystal changes in dependence on SOC, temperature of the negative half cell electrolyte is also shown